Passionfruit review

Strong — 4 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 25% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Strong 8 · Adequate 7 · Weak 1. The typical agency here scores Strong, and none scores higher than this one.

What this dimension measures: What the firm actually does, described concretely enough to evaluate: content restructuring for extractability, entity and knowledge-graph work, schema, source and citation-footprint building, presence on the corpora assistants actually draw from, llms.txt and crawler access. Be sceptical of anything indistinguishable from ordinary SEO sold at a premium under a new name: if the described method is only 'good content and schema', say so plainly in the verdict rather than crediting the label.

Scores high — Named, specific techniques tied to how retrieval works score 4-5.

Scores low — 'We optimise for AI' with no described mechanism scores 1-2.

What we found — Named, retrieval-specific techniques rather than 'we optimise for AI'. The crawler post names the exact retrieval user agents to allow (OAI-SearchBot, Claude-SearchBot, PerplexityBot, ChatGPT-User, Claude-User, Perplexity-User), separates retrieval bots from training crawlers as a distinct decision, and tells readers to filter server logs on those tokens and check for 403/429 responses and to verify bot identity against the crawler IP JSON files OpenAI, Anthropic and Perplexity publish. The citation-source post describes building a citation footprint by classifying cited URLs into brand-owned, earned media, community, reference and video. The Visily case study describes 40-50 word definitions placed immediately after the H1 'formatted for extraction' plus JSON-LD, speakable and SoftwareApplication schema. Capped below Excellent: llms.txt, entity resolution and knowledge-graph work are absent from what is published, and a large share of the delivered work described in case studies is ordinary technical and content SEO (canonicalization of 18 brand-name variants, content architecture, Lighthouse 96/100, 73% JavaScript execution reduction) rather than anything specific to generative retrieval. That is the high band above, which is why it scored Strong.

On the record — “Concrete crawler-access method is published, naming the retrieval user agents to allow (OAI-SearchBot, Claude-SearchBot, PerplexityBot, ChatGPT-User, Claude-User, Perplexity-User), separating them from training crawlers, and directing verification via server-log user-agent filtering, 403/429 checks and the crawler IP range files published by OpenAI, Anthropic and Perplexity.” getpassionfruit.com ↗

Evidence: partly checkable — corroborated in part against the sources below; the remainder rests on the agency’s own account.

Excellent — 5 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 20% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Strong 6 · Adequate 9 · Weak 1. The typical agency here scores Adequate, and none scores higher than this one.

What this dimension measures: How the firm proposes to prove anything, in a setting where outputs are non-deterministic and vary by user, session and region. Look for a described tracking approach with its limits acknowledged: prompt sets, sampling frequency, variance handling, and how a citation is attributed.

Scores high — A stated method that names non-determinism as a limit scores 4-5.

Scores low — Vague 'we track your AI visibility' scores 2-3. A firm presenting single-prompt screenshots as measurement scores 1-2 — one answer on one day is an anecdote, and treating it as a metric is the characteristic error of this category.

What we found — The strongest thing on this site, and unusually explicit for the category. Their published research states that a single-run result is not a measurement: 'That #4 ranking your vendor reported was a snapshot of a single run of a single prompt on a single platform. It is not a position. It is not a trend.' It quantifies the non-determinism ('The odds of getting the same brand list twice from ChatGPT? Less than 1 in 100'), sets a sampling floor ('visibility percentage (how often you appear across 60-100+ runs) is the only citation metric with statistical validity. If your vendor runs each prompt fewer than 60 times before reporting, the data is not meaningful'), and states the attribution limit plainly: 'No study has yet demonstrated a reliable causal link between citation metrics and revenue or pipeline outcomes.' The citation-source post specifies a working prompt set of 20-40 prompts across recommendation, comparison and evaluation intents, warns that 'fewer than 15 prompts tend to surface noise rather than signal', prescribes weekly runs for at least three weeks before drawing conclusions, and gives a decay figure of 40-60% source rotation month over month. This is a stated method that names non-determinism as a limit rather than hiding it. That is the high band above, which is why it scored Excellent.

On the record — “Published research sets an explicit statistical floor for AI visibility measurement: 'visibility percentage (how often you appear across 60-100+ runs) is the only citation metric with statistical validity. If your vendor runs each prompt fewer than 60 times before reporting, the data is not meaningful.'” getpassionfruit.com ↗

On the record — “The same research states plainly that the outcome link is unproven: 'No study has yet demonstrated a reliable causal link between citation metrics and revenue or pipeline outcomes.'” getpassionfruit.com ↗

attribution — working out which marketing touch actually caused a sale. Good practice names its model and its blind spots; bad practice quotes each ad platform’s self-graded numbers, which overlap and overclaim.

pipeline — the dollar value of open, qualified deals sales is working. “Pipeline created” ties marketing to money, where lead counts don’t.

Evidence: verified — checked against a named source you can open; the links under Sources below are where to check it yourself.

Adequate — 3 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 20% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Strong 5 · Adequate 7 · Weak 4. The typical agency here scores Adequate, and 5 of them score higher than this one.

What this dimension measures: Weighted unusually high because this category's defining risk is overclaim, not opacity. No vendor controls what an assistant says. Also flag results claims presented without the prompt set or timeframe that produced them, and any implication that the firm has privileged access to a model's ranking.

Scores high — A firm that states plainly what it cannot guarantee scores 4-5.

Scores low — Guaranteed rankings, guaranteed citations, guaranteed 'first result in ChatGPT', or claimed partnerships with model providers score 1-2 and must be named in red_flags.

What we found — No guarantee language was found on any page read: no guaranteed rankings, no guaranteed citations, no 'first result in ChatGPT', and no claimed partnership with any model provider (ChatGPT, Perplexity, Claude, Gemini, AI Overviews and Copilot are described as channels they optimise into, not relationships they hold). That clears the category's worst failure modes. Held to Adequate because the marketing surface does not follow the standard their own research sets. The homepage advertises '750% AI visibility growth' and '10.5M AI Search Traffic', and the case-study index carries '+5,237% AI Revenue (YoY)', '59x AI Revenue (YoY)' and '+23% AI Revenue in 30 Days' with no prompt set, no attribution method, no tool named and no baseline - precisely the presentation their own research calls not meaningful. Separately, they publish a '10 Best Ecommerce SEO Agencies 2026' ranking on their own domain that places Passionfruit at #1 and states in its FAQ that 'Passionfruit ranks as the top ecommerce SEO agency for 2026', without disclosing at the outset that the publisher is the top-ranked entry. That is between the two bands, which is why it scored Adequate.

On the record — “No guaranteed ranking, guaranteed citation, or claimed partnership with a model provider appears on any page read. Model surfaces (ChatGPT, Perplexity, Claude, Gemini, AI Overviews, Copilot) are described only as channels optimised into.” getpassionfruit.com ↗

On the record — “The firm publishes a '10 Best Ecommerce SEO Agencies 2026' ranking on its own domain in which it ranks itself #1, with an FAQ stating 'Passionfruit ranks as the top ecommerce SEO agency for 2026', without disclosing the conflict at the outset.” getpassionfruit.com ↗

attribution — working out which marketing touch actually caused a sale. Good practice names its model and its blind spots; bad practice quotes each ad platform’s self-graded numbers, which overlap and overclaim.

Evidence: verified — checked against a named source you can open; the links under Sources below are where to check it yourself.

Strong — 4 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 15% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Strong 9 · Adequate 7. The typical agency here scores Strong, and none scores higher than this one.

What this dimension measures: Attributable client work. Be lenient on volume and strict on attribution: this category is young enough that a short roster is expected, but a long roster of unnamed logos is not evidence. Where a firm's OWN brand is demonstrably visible in AI answers, note it as a real if partial signal — it is at least a claim about themselves they cannot make privately.

Scores low — Named clients with described engagements score 4-5; anonymised or absent scores 2-3.

What we found — Real named attribution, read first-hand. The homepage names Visily, Necesera, Camb AI, Vitality, EUME, Typsy Beauty and Powersutra, and named case studies exist at their own URLs with described engagements and dated windows: 'How Visily Grew 2M+ Impressions and 146 SERP rankings from Scratch' (approx. April/May 2025 to October 2025, 28,440 keywords ranked, 261 People Also Ask boxes, 47 AI Overview positions) and 'NeceSera's Journey to 3x Clicks and 4.14M Impressions In 6 Months' (April 2024 to May 2025, clicks 14,800 to 60,500, impressions 1.13M to 4.36M, average position 17.0 to 14.5). Capped at Strong: the /case-studies index itself is mostly anonymised by vertical ('Kitchen Appliances Brand', 'Luxury Womenswear', 'Cybersecurity Platform'), the testimonials page carried no attributable name, role and company in the content returned, and none of the engagements could be corroborated away from the vendor's own site.

On the record — “Named client case studies with dated windows and before/after figures are published, including Visily (approx. April/May 2025 to October 2025; 2M+ impressions, 146 SERP feature rankings, 28,440 keywords) and NeceSera (April 2024 to May 2025; clicks 14,800 to 60,500, impressions 1.13M to 4.36M, average position 17.0 to 14.5).” getpassionfruit.com ↗

attribution — working out which marketing touch actually caused a sale. Good practice names its model and its blind spots; bad practice quotes each ad platform’s self-graded numbers, which overlap and overclaim.

Evidence: partly checkable — corroborated in part against the sources below; the remainder rests on the agency’s own account.

Weak — 2 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 10% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Excellent 3 · Strong 4 · Adequate 3 · Weak 6. The typical agency here scores Adequate, and 10 of them score higher than this one.

What this dimension measures: Note contract length: long lock-ins are hard to justify in a category this young, where nobody can evidence a twelve-month outcome yet, and a firm offering short or month-to-month terms should be credited for it.

Scores low — Published prices, ranges or defined packages score 4-5; ranges on request score 2-3; bespoke-only scores 1-2.

What we found — The /pricing page exists but publishes no number. It says 'Custom Pricing' and 'Tailored to your needs', and routes every buyer to an exploratory call and a personalised audit. No price, no range, no package tiers, no minimum term and no contract length are stated anywhere read, so a buyer cannot tell what an engagement costs or how long they would be committed - a particular problem in a category where nobody can yet evidence a twelve-month outcome. Kept above Poor only because a defined pre-sale process is published and they state they may recommend against their full services where these are unnecessary.

On the record — “No price, range, package tier or contract length is published. The pricing page states only 'Custom Pricing' and 'Tailored to your needs', routing all buyers to an exploratory call and a personalised audit.” getpassionfruit.com ↗

Evidence: verified — checked against a named source you can open; the links under Sources below are where to check it yourself.

Stronger here: Cubitrek scores Excellent on the same dimension.

Strong — 4 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 5% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Strong 6 · Adequate 10. The typical agency here scores Adequate, and none scores higher than this one.

What this dimension measures: Whether the firm acknowledges that the surfaces move, and describes how it adapts: monitoring for model and interface changes, revisiting method, published research or original testing.

Scores high — Evidence of ongoing testing and updated method scores 4-5.

Scores low — A static playbook presented as settled practice scores 2, since nothing in this category has been stable long enough to be settled.

What we found — Genuine original testing rather than a static playbook. A research section carries eight pieces, including a first-hand study scoring roughly 350 top-10 ranking URLs across 35 SERPs spanning seven intent categories, using an eight-dimension 0-24 manual scale and explicitly rejecting classifiers: 'We did not use any detection tool. We scored what was visible in the text because classifiers are noisy.' They also track surface movement over time, citing ChatGPT's Reddit citation share falling from nearly 60% of responses in early August 2025 to around 10% by mid-September. Held below Excellent because individual research pieces carried no visible publication dates in the content returned, so how current the testing is cannot be confirmed, and the 350-page study states no limitations. That is the high band above, which is why it scored Strong.

On the record — “Original first-hand testing exists: roughly 350 top-10 ranking URLs scored across 35 SERPs on an eight-dimension 0-24 manual scale, with the authors stating 'We did not use any detection tool.'” getpassionfruit.com ↗

Evidence: partly checkable — corroborated in part against the sources below; the remainder rests on the agency’s own account.

Weak — 2 of 5 on this rubric’s scale, from Poor (1) to Excellent (5). This dimension carries 5% of the total score.

Benchmark — across the 16 other agencies evaluated in this discipline, this dimension runs Excellent 2 · Strong 6 · Adequate 6 · Weak 2. The typical agency here scores Strong, and 14 of them score higher than this one.

What this dimension measures: Independent, verified reviews. Low weight and low expectations: most firms here are under two years old, so an absent review base is a fact about the category's age rather than a finding about the vendor.

Scores low — A substantial verified base scores 4-5; a handful scores 2-3; none located scores 2.

What we found — No independent review base could be read first-hand. A G2 profile exists for 'Passionfruit Labs' (their tracking product, not the agency service), but g2.com returned HTTP 403 to direct retrieval, so no rating or count is cited here. The Clutch profile surfaced by search - 'Passionfruit Creative' - is a different company. Per the rubric this is a low-weight, low-expectation dimension and an absent readable base is largely a fact about the category's age.

Evidence: inferred — our reading of indirect evidence, not a documented fact.

Stronger here: Intero Digital scores Excellent on the same dimension.